Proton vs photon radiotherapy — normal-tissue & growth-plate sparing in a child's brain tumor
Radiotherapy in children is not simply "adult radiotherapy at a smaller scale." A child irradiated at age 5 may live another 75+ years — decades in which any radiation-induced injury to a still-growing organ has time to fully express itself. This simulator models a common, curable pediatric brain tumor to show why normal-tissue sparing matters as much as tumor control.
Two facts distinguish pediatric radiotherapy from adult practice. First, life-years at risk: a cured 6-year-old has a life expectancy that stretches many decades past the point where secondary cancers, endocrine failure, or cognitive decline typically manifest — effects that an adult treated at 65 will often not live long enough to experience. Second, developing tissue is more radiosensitive: growth plates, myelinating white matter, the cochlear hair cells and stria vascularis, and the hypothalamic-pituitary axis are all still undergoing active proliferation and maturation during the treatment years, and radiation preferentially disrupts dividing and differentiating cells.
These two facts compound: the same physical dose (Gy) causes more biological injury in a child, and that injury then has more time to accumulate clinical consequences over a normal lifespan.
A framework often used in pediatric oncology: cure is necessary but not sufficient. A treatment plan is judged on its combination of tumor control probability AND normal tissue complication probability, weighted over 70+ remaining years — not just 5-year survival.
Medulloblastoma, the most common malignant pediatric brain tumor, arises in the cerebellum (posterior fossa) and can seed through the cerebrospinal fluid to coat the entire brain and spinal cord surface. Standard therapy for most patients therefore treats the whole craniospinal axis (CSI) to a modest dose (23.4–36 Gy depending on risk group), followed by a boost to the primary tumor bed (total ~54–55.8 Gy).
This creates a uniquely large treatment volume for a pediatric case: essentially the entire central nervous system, plus the vertebral bodies that house it, must receive meaningful dose — which is exactly why the entrance/exit-dose behavior of the chosen radiation modality matters so much here.
• Cochlea (yellow) — hair cells and stria vascularis sit near the posterior fossa boost field and are exquisitely sensitive to radiation, especially combined with platinum-based chemotherapy. • Pituitary / hypothalamus (purple) — the master endocrine axis controlling growth hormone, thyroid, and puberty; it sits centrally and receives dose from any whole-brain or skull-base field. • Temporal lobes / whole brain (blue) — white-matter and cortical volume whose dose correlates with neurocognitive and IQ trajectory over years of follow-up. • Vertebral growth plates (green) — the epiphyseal plates at each vertebral body that drive spinal column growth; if a spinal photon field exits through the vertebral body, these plates receive nearly the full prescription dose.
Modern photon planning (IMRT or VMAT) is very good at conforming the high-dose region to the target. But every photon beam obeys exponential attenuation: it deposits dose on the way in, through the target, and on the way out the other side. Stack five to nine such beams around a patient and the low-dose "leftovers" from each one overlap across a large fraction of the body — the low-dose bath.
A megavoltage photon beam deposits its maximum dose a few centimeters below the skin, then decays roughly exponentially with depth, continuing to deposit non-trivial dose all the way through and out the far side of the patient. IMRT/VMAT achieves target conformality by intensity-modulating and combining many such beams from different angles, so that only the target receives dose from every beam simultaneously — but tissue anywhere along any single beam's path still receives some dose from that beam.
Summed across all beam angles, this produces a large volume of normal tissue that receives a low-to-moderate dose even though it is nowhere near the tumor. This is captured by the physics concept of "integral dose" — the total energy deposited in the patient, summed over all tissue, not just the target.
Craniospinal irradiation is a worst case for the photon low-dose bath because the target already spans nearly the entire CNS. A whole-brain field necessarily also irradiates temporal lobes, the hypothalamic-pituitary axis, and both cochleae with near-prescription dose, since they lie inside or immediately adjacent to the treated brain volume. The spinal field, delivered from a posterior approach, must exit anteriorly — passing through the vertebral bodies and, depending on level, the heart, lungs, thyroid, or abdominal organs before leaving the patient. Anterior vertebral body dose from this exit path is a principal cause of growth-plate injury and disproportionate short stature in CSI survivors.
Multiple cooperative-group studies correlate whole-brain / temporal lobe photon dose with measurable full-scale IQ decline over the years following treatment — one of the best-established late effects in pediatric neuro-oncology.
Radiation-induced secondary malignancy risk is modeled (via the linear no-threshold hypothesis used in radiation protection) as roughly proportional to the integral dose received by normal tissue, integrated over the volume of tissue exposed and the years of life remaining for that risk to manifest. Because a photon plan deposits dose broadly (the bath) rather than only within the target, it maximizes exactly the quantity — normal-tissue integral dose — that drives this risk. This effect is disproportionately important in children, whose growing, actively dividing tissue is more radiosensitive per unit dose and who have far more remaining years for a radiation-induced tumor to develop.
Protons behave completely differently in tissue. A proton beam deposits relatively little energy while traveling at high speed, then — as it slows near the end of its range — deposits a sharp spike of dose called the Bragg peak, before stopping almost completely. Modern pencil-beam scanning (PBS) exploits this by "painting" the target one small spot and one energy layer at a time.
Unlike photons, charged protons have a finite, energy-defined range in tissue. As a proton beam slows down, the rate of energy loss per unit path length increases sharply (described by the Bethe-Bloch relationship), producing a narrow, intense peak of dose deposition — the Bragg peak — right before the proton stops entirely. Beyond that depth, dose falls essentially to zero.
A single Bragg peak is too narrow to cover a real tumor, so pencil-beam scanning delivers many individual "spots" of varying energy (depth) and lateral position, stacking Bragg peaks to build a Spread-Out Bragg Peak (SOBP) that conforms to the target's full 3D shape — while still dropping to near-zero dose just beyond the target's distal edge.
Because the beam simply stops, there is no equivalent of photon exit dose. A posterior spinal proton field delivers its entrance dose to skin and posterior elements, rises through the vertebral body, and then stops inside it — sparing the heart, lungs, thyroid, and abdominal organs that a photon beam would have to exit through.
For the same medulloblastoma craniospinal case, a proton plan can treat the posterior fossa boost and the entire spinal axis using beams that stop at or just beyond the target depth. The anterior vertebral body — and the growth plates within it — sits largely beyond the proton range and receives only a fraction of the dose a photon exit beam would deliver. Likewise, a posterior fossa boost delivered with protons can spare much of the supratentorial brain, temporal lobes, and cochleae that a photon whole-brain or boost field would otherwise dose more broadly. Proton spinal fields also avoid the beam-matching junction problems that arise when multiple photon fields must be abutted along a growing spine.
Prospective cohort studies of proton craniospinal irradiation in medulloblastoma (e.g., Yock et al., Lancet Oncology, 2016) report stable IQ and neurocognitive trajectories over years of follow-up, in contrast to the progressive declines historically reported after photon whole-brain / CSI treatment. Dosimetric comparison studies consistently show substantially lower mean dose to the cochlea, hypothalamic-pituitary axis, and vertebral bodies with proton plans versus photon plans of equivalent target coverage, and several series report lower rates of clinically significant hearing loss requiring hearing aids. These findings are the primary clinical rationale for proton therapy being preferentially recommended for pediatric CNS irradiation where access allows.
Aggregate "integral dose" is useful, but clinical late effects are driven by dose to specific organs at risk. This stage compares modeled mean dose to four structures under equivalent photon and proton craniospinal plans — the gap between the two bars for each structure is the sparing proton therapy is expected to provide.
Sensorineural hearing loss after CSI is driven by cumulative dose to the cochlea, and the risk rises sharply above a mean cochlear dose threshold in the low-to-mid 30s Gy — right in the range a photon posterior fossa boost typically delivers, since the cochleae sit adjacent to the boost volume. Hearing loss risk is further amplified by concurrent platinum-based chemotherapy (cisplatin), which is ototoxic on its own and synergizes with radiation injury to the hair cells and stria vascularis. Lowering cochlear dose is therefore one of the most direct, well-quantified benefits of proton planning in this disease.
The hypothalamic-pituitary axis controls growth hormone, thyroid-stimulating hormone, and the gonadotropins that drive puberty; its dose-response for endocrinopathy is graded, with growth hormone deficiency occurring at comparatively low doses and full panhypopituitarism at higher doses. Because it sits centrally in the brain, any whole-brain or skull-base field doses it heavily.
Vertebral growth plates are a separate but related concern: when a photon spinal field exits anteriorly through the vertebral body, the growth plates there receive near-prescription dose, producing vertebral body growth arrest, disproportionate trunk shortening relative to limb length, and a higher risk of scoliosis as the child grows asymmetrically. Sparing the anterior vertebral body — which the proton beam's finite range achieves almost automatically — directly addresses this.
Temporal lobe and whole-brain dose correlate with measurable declines in processing speed, working memory, and full-scale IQ over the years following CSI, with younger age at treatment associated with steeper decline. Separately, integral dose to normal tissue — summed across the whole body, not any one organ — is the principal driver of radiation-induced secondary malignancy risk, and this risk is proportionally higher in pediatric patients because their tissue is more radiosensitive per unit dose and they have decades of remaining life for a second tumor (e.g., meningioma, glioma, sarcoma) to develop. Both endpoints track closely with the volume of normal tissue that leaves the treatment field having received meaningful dose — precisely the quantity a proton plan minimizes.
This final stage translates the dose reductions above into simplified, illustrative risk-reduction bars across four outcome domains. These are qualitative, modeled projections built from published dose-response associations in the literature — they are not a validated predictive nomogram and should never be read as an individual patient's personalized risk estimate.
Each bar in this stage combines two inputs already shown in this simulator: (1) the modeled dose reduction to the relevant structure(s) from Stage 4, and (2) a qualitative sense, drawn from published cohort studies, of how strongly that structure's dose correlates with the named outcome. The result is a relative risk-reduction index, not an absolute probability — it says "meaningfully lower," not "reduced from X% to Y% for your child." Real outcome prediction requires an individualized dosimetric plan, the patient's specific chemotherapy regimen, genetic factors, and validated clinical nomograms, none of which this illustrative tool has access to.
Read every bar in this stage as directional and illustrative. It is built to communicate why normal-tissue sparing matters, not to forecast any individual outcome — that judgment belongs to a pediatric oncology and radiation oncology team.
• Hearing preservation — lower cochlear dose is modeled as a large relative benefit, consistent with the well-characterized cochlear dose-response threshold for sensorineural hearing loss. • Growth / endocrine axis — lower hypothalamic-pituitary dose is modeled as a moderate-to-large benefit, reflecting the graded dose-response for growth hormone deficiency and other pituitary hormone axes. • Neurocognitive / IQ trajectory — lower whole-brain and temporal lobe dose is modeled as a moderate benefit, reflecting a real but multifactorial relationship (chemotherapy, surgery, hydrocephalus, and age at treatment all also contribute). • Secondary malignancy — lower integral dose is modeled as a moderate-to-large benefit, reflecting the linear no-threshold assumption linking normal-tissue dose to long-term second-cancer risk.
Proton benefits shown here come with real-world trade-offs. Young children often cannot hold still for the minutes-long daily setup precision proton (and modern photon) treatment requires, so children under roughly 7 years old frequently need daily general anesthesia for the multi-week course — itself a nontrivial cumulative exposure and logistical burden. Proton centers remain far less numerous than photon-capable linear accelerators, so geographic access, treatment-related travel, and insurance authorization can be significant barriers. Motion and anatomical changes during treatment (weight change, positioning variability) also require careful image-guidance and plan-robustness management, since the sharp Bragg peak that gives protons their sparing advantage is also more sensitive to range uncertainty than a photon beam. None of these caveats change the underlying physics advantage — they are the practical reasons access and workflow, not efficacy, are often the limiting factor.